Fuzzy Swarm Net (FSN) for Classification in Data Mining

Fuzzy Swarm Net (FSN) for Classification in Data Mining

Abstract

In this paper, we have used a fuzzy net trained by the particle swarm optimization (PSO) technique for classification task of data mining. Fuzzy net uses a single layer neural network to reduce the complexities associated with multi-layer perceptron (MLP). In addition, we trained the network by using PSO to obtain optimal weight vectors, which reduces the problem of being trapped into a local minimum. From the simulation study, it is observed that the fuzzy swarm net (FSN) gives a promising result. Further, we compared the result obtained from FSN with multi-layer perceptron (MLP) radial basis function (RBF). The results obtained from FSN draws a clear edge between FSN and RBF and a competitive result with MLP.

References